You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@hbase.apache.org by "Allan Yang (JIRA)" <ji...@apache.org> on 2019/03/19 03:17:00 UTC

[jira] [Commented] (HBASE-20951) Ratis LogService backed WALs

    [ https://issues.apache.org/jira/browse/HBASE-20951?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16795593#comment-16795593 ] 

Allan Yang commented on HBASE-20951:
------------------------------------

[~elserj], I have a question for the ratis based log service. The goal of this project is to separate WAL from HDFS and store it into a log service to achieve lower cost and flexibility in cloud compare to HDFS(decommission is panic as said in the doc). But, what is the storage for this log service? Local disks in EC2 instance? How to achieve the goal of flexibility itself?Does it mean that we need to maintain another cluster for this log service? isn't it panic too when decommissioning? 

> Ratis LogService backed WALs
> ----------------------------
>
>                 Key: HBASE-20951
>                 URL: https://issues.apache.org/jira/browse/HBASE-20951
>             Project: HBase
>          Issue Type: New Feature
>          Components: wal
>            Reporter: Josh Elser
>            Assignee: Josh Elser
>            Priority: Major
>
> Umbrella issue for the Ratis+WAL work:
> Design doc: [https://docs.google.com/document/d/1Su5py_T5Ytfh9RoTTX2s20KbSJwBHVxbO7ge5ORqbCk/edit#|https://docs.google.com/document/d/1Su5py_T5Ytfh9RoTTX2s20KbSJwBHVxbO7ge5ORqbCk/edit]
> The (over-simplified) goal is to re-think the current WAL APIs we have now, ensure that they are de-coupled from the notion of being backed by HDFS, swap the current implementations over to the new API, and then wire up the Ratis LogService to the new WAL API.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)